Tools are provided for (i) obtaining image data in which information is embedded by printing multiple types of patterns in a matrix, (ii) calculating multiple vertical-line-segment sequences and multiple horizontal-line-segment sequences, obtained by connecting vertically arranged patterns and horizontally arranged patterns, respectively, in the image data with polygonal lines, (iii) searching for a vertical line segment and a horizontal line segment intersecting with each other for each combination of one of the vertical-line-segment sequences and one of the horizontal-line-segment sequences, (iv) calculating an intersection point between the thus-found vertical line segment and the horizontal line segment as a position of a pattern, and (v) extracting the information embedded in the printed material by identifying types of patterns at the calculated intersection points.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. An information extraction apparatus comprising: an obtaining unit that obtains image data that includes therein information that are printed with multiple types of patterns in a matrix therein; a line-segment-sequence calculating unit that calculates a plurality of vertical-line-segment sequences, each of which is obtained by connecting patterns arranged in a vertical row in the image data with a particular polygonal line, and calculates a plurality of horizontal-line-segment sequences, each of which is obtained by connecting patterns arranged in a horizontal row in the image data with another polygonal line; a search unit that searches, for each combination of one vertical-line-segment sequence of the vertical-line-segment sequences and one horizontal-line-segment sequence of the horizontal-line-segment sequences, vertical line segments that form the one vertical-line-segment sequence and horizontal line segments that form the one horizontal-line-segment sequence for a vertical line segment and a horizontal line segment that intersect with each other; an intersection-point calculating unit that calculates a point of intersection between the vertical line segment and the horizontal line segment having been found by the search unit as a position of a pattern; and an extracting unit that extracts the information by identifying types of patterns at the points, calculated by the intersection-point calculating unit, of intersection between the vertical line segments and the horizontal line segments, wherein the search unit selects a vertical line segment and a horizontal line segment to be subjected to intersection determination calculates a point of intersection between the selected vertical line segment or an extension thereof and the selected horizontal line segment or an extension thereof by using values of parameters, and determines whether the vertical line segment and the horizontal line segment intersect with each other based on the values of the parameters, and wherein if the vertical line segment and the horizontal line segment are determined not to be intersecting with each other based on the values of the parameters the search unit selects a vertical line segment and a horizontal line segment to be subjected to intersection determination next based on the values of the parameters.
An information extraction system reads an image containing a matrix of printed patterns representing data. It finds vertical and horizontal lines by connecting the patterns in rows and columns with polygonal lines. The system then finds the intersections of vertical and horizontal line segments. For each pair of vertical and horizontal line sequences, it checks if individual line segments intersect. Intersection is determined by calculating a potential intersection point using parameter values and checking if it lies on both segments (or their extensions). If no intersection is found for a pair of segments, it selects the next vertical and horizontal line segments and repeats the intersection check based on parameter values. Finally, it identifies the type of pattern at each intersection point to extract the embedded information.
2. The information extraction apparatus according to claim 1 , wherein the search unit selects, based on the values of the parameters, the vertical line segment and the horizontal line segment to be subjected to intersection determination next from among vertical line segments connected to the vertical line segment determined not to be intersecting and from among horizontal line segments connected to the horizontal line segment determined not to be intersecting.
In the information extraction system, as described above, the system uses parameter values to select the next vertical and horizontal line segments to test for intersection. Specifically, if a vertical and horizontal line segment pair do not intersect, the system chooses the next segments from those connected to the non-intersecting segments. This focuses the search on segments likely to intersect, based on their proximity to previously checked segments.
3. The information extraction apparatus according to claim 1 , wherein the search unit defines a vertical-line-segment partial sequence that is a line segment sequence that possibly intersects with a horizontal line segment, defines a horizontal-line-segment partial sequence that is a line segment sequence that possibly intersects with a vertical line segment, and selects the vertical line segment and the horizontal line segment to be subjected to intersection determination next from the vertical-line-segment partial sequence and the horizontal-line-segment partial sequence based on the values of the parameters.
In the information extraction system, as described above, the system focuses the search for intersecting line segments by defining "partial sequences." A vertical-line-segment partial sequence contains line segments likely to intersect a horizontal line segment, and a horizontal-line-segment partial sequence contains segments likely to intersect a vertical segment. The system then selects the next vertical and horizontal line segments to test for intersection from within these partial sequences, based on parameter values, narrowing the search space.
4. The information extraction apparatus according to claim 3 , wherein the search unit selects, as the vertical line segment and the horizontal line segment to be subjected to intersection determination next, a vertical line segment that is positioned at a substantial center of the vertical-line-segment partial sequence and a horizontal line segment that is positioned at a substantial center of the horizontal-line-segment partial sequence.
In the information extraction system with partial sequences, as described above, the system selects the vertical and horizontal line segments to test for intersection from the approximate center of their respective partial sequences. It chooses the line segment in the vertical partial sequence closest to the middle and the line segment in the horizontal partial sequence closest to the middle. This assumes that intersections are most likely to occur near the center of the relevant sequences.
5. The information extraction apparatus according to claim 1 , wherein the search unit selects, as a vertical line segment and a horizontal line segment to be subjected to intersection determination initially, a vertical line segment that is positioned at a substantial center of the vertical-line-segment sequence and a horizontal line segment that is positioned at a substantial center of the horizontal-line-segment sequence.
In the information extraction system, as described above, the initial vertical and horizontal line segments to test for intersection are chosen from the approximate centers of the overall vertical and horizontal line segment sequences. The system selects the line segment in the vertical sequence closest to the middle and the line segment in the horizontal sequence closest to the middle as starting points for the intersection search.
6. The information extraction apparatus according to claim 1 , wherein each of the vertical-line-segment sequences is a sequence of line segments obtained by connecting patterns arranged in a vertical row in the image data with a particular polygonal line in an order of coordinate values that define vertical positions, and each of the horizontal-line-segment sequences is a sequence of line segments obtained by connecting patterns arranged in a horizontal row in the image data with another polygonal line in an order of coordinate values that define horizontal positions.
In the information extraction system, as described above, the vertical line segment sequences are created by connecting patterns in a vertical row in ascending order of their vertical coordinates. Similarly, horizontal line segment sequences are created by connecting patterns in a horizontal row in ascending order of their horizontal coordinates. This ensures the segments are ordered logically along each row or column.
7. The information extraction apparatus according to claim 1 , wherein the obtaining unit controls a reading unit to read the printed material to obtain image data.
In the information extraction system, as described above, the system obtains the image data by controlling a reading unit (like a scanner or camera) to capture an image of the printed material. The reading unit is an integral part of the overall information extraction apparatus.
8. The information extraction apparatus according to claim 1 , wherein the obtaining unit obtains the image data via a network from a reading apparatus that reads the printed material to obtain the image data.
In the information extraction system, as described above, the system obtains the image data from a remote reading apparatus over a network. The remote apparatus reads the printed material and sends the image data to the information extraction system for processing.
9. An information extraction method comprising: (a) obtaining image data that includes therein information that are printed with multiple types of patterns in a matrix therein, the obtaining being performed by an obtaining unit; (b) calculating a plurality of vertical-line-segment sequences, each of which is obtained by connecting patterns arranged in a vertical row in the image data with a particular polygonal line, and calculating a plurality of horizontal-line-segment sequences, each of which is obtained by connecting patterns arranged in a horizontal row in the image data with another polygonal line, the calculating being performed by a line-segment-sequence calculating unit; (c) searching, for each combination of one vertical-line-segment sequence of the vertical-line-segment sequences and one horizontal-line-segment sequence of the horizontal-line-segment sequences, vertical line segments that form the one vertical-line-segment sequence and horizontal line segments that form the one horizontal-line-segment sequence for a vertical line segment and a horizontal line segment that intersect with each other, the searching being performed by a search unit; (d) calculating a point of intersection between the vertical line segment and the horizontal line segment having been found by the search unit as a position of a pattern, the calculating being performed by an intersection-point calculating unit; and (e) extracting the information by identifying types of patterns at the points, calculated by the intersection-point calculating unit, of intersection between the vertical line segments and the horizontal line segments, the extracting being performed by an extracting unit, wherein in (c), the search unit selects a vertical line segment and a horizontal line segment to be subjected to intersection determination, calculates a point of intersection between the selected vertical line segment or an extension thereof and the selected horizontal line segment or an extension thereof by using values of parameters, and determines whether the vertical line segment and the horizontal line segment intersect with each other based on the values of the parameters, and if the vertical line segment and the horizontal line segment are determined not to be intersecting with each other based on the values of the parameters, the search unit selects a vertical line segment and a horizontal line segment to be subjected to intersection determination next based on the values of the parameters.
An information extraction method involves these steps: (a) obtaining image data containing a matrix of patterns representing information; (b) calculating vertical and horizontal line segment sequences by connecting patterns in rows and columns with polygonal lines; (c) searching for intersecting vertical and horizontal line segments from corresponding line sequences; (d) calculating intersection points between the found segments to locate the patterns; (e) extracting information by identifying pattern types at the intersection points. The intersection search involves selecting a vertical and horizontal line segment pair, calculating their intersection point (or extensions), and checking for actual intersection using parameter values. If no intersection, the next segment pair is selected based on parameter values, and the process repeats.
10. An information-extraction computer program product comprising a non-transitory computer-readable storage medium having computer-readable program codes embodied in the storage medium, the program codes, when executed, causing a computer to perform an information extraction method comprising: (a) obtaining image data that includes therein information that are printed with multiple types of patterns in a matrix therein; (b) calculating a plurality of vertical-line-segment sequences, each of which is obtained by connecting patterns arranged in a vertical row in the image data with a particular polygonal line, and calculating a plurality of horizontal-line-segment sequences, each of which is obtained by connecting patterns arranged in a horizontal row in the image data with another polygonal line; (c) searching, for each combination of one vertical-line-segment sequence of the vertical-line-segment sequences and one horizontal-line-segment sequence of the horizontal-line-segment sequences, vertical line segments that form the one vertical-line-segment sequence and horizontal line segments that form the one horizontal-line-segment sequence for a vertical line segment and a horizontal line segment that intersect with each other; (d) calculating a point of intersection between the vertical line segment and the horizontal line segment having been found by the search unit as a position of a pattern; and (e) extracting the information by identifying types of patterns at the points, calculated by the intersection-point calculating unit, of intersection between the vertical line segments and the horizontal line segments, wherein (c) is performed by a search unit, and the search unit selects a vertical line segment and a horizontal line segment to be subjected to intersection determination, calculates a point of intersection between the selected vertical line segment or an extension thereof and the selected horizontal line segment or an extension thereof by using values of parameters, and determines whether the vertical line segment and the horizontal line segment intersect with each other based on the values of the parameters, and if the vertical line segment and the horizontal line segment are determined not to be intersecting with each other based on the values of the parameters, the search unit selects a vertical line segment and a horizontal line segment to be subjected to intersection determination next based on the values of the parameters.
A computer program product containing instructions, when executed, performs an information extraction method. The method comprises: (a) obtaining image data containing a matrix of printed patterns representing information; (b) calculating vertical and horizontal line segment sequences by connecting patterns in rows and columns with polygonal lines; (c) searching for intersecting vertical and horizontal line segments from corresponding line sequences; (d) calculating intersection points between the found segments to locate the patterns; (e) extracting information by identifying pattern types at the intersection points. The intersection search involves selecting a vertical and horizontal line segment pair, calculating their intersection point (or extensions), and checking for actual intersection using parameter values. If no intersection, the next segment pair is selected based on parameter values, and the process repeats.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
May 13, 2010
August 20, 2013
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.